Demystifying Data

04 - The Power and Perils of Statistics

2023-12-14

Statistics

Origins

19th Century Industrial Nations

Demographics


Population


Industrial Output

Sample vs Population

Probability Theory

Gambling / Casinos

Meyer Lansky


Guinness


Student-t

Mark Twain


There are three kinds of lies: …

… lies,

… damned lies,

… and statistics.

Misleading Statistics

How to Lie with Statistics


Correlation / Causation

Post Hoc, Ergo Proptor Hoc

Replication Crisis


Relative Variation

Bad Visualisation

Pie Charts Considered Harmful

What to Do?

Quantify Uncertainty

Not-Bad DataViz


Lines


Points


Bars

Model Validation

Test / Train Datasets


Cross-Validation

Summary

Thank You


mcooney@describedata.com


https://kaybenleroll.github.io/data_workshops/talk_cirdas_master_202311/